Netris raises $15M Series A from a16z to accelerate network provisioning for AI neoclouds

What Happened

Netris has secured a $15 million Series A funding round led by Andreessen Horowitz (a16z). The company develops software that runs directly on bare-metal network switches, providing a cloud-like consumption model for physical networking infrastructure. This funding is explicitly aimed at accelerating the deployment of "neoclouds"—emerging, GPU-heavy cloud providers that cater specifically to intensive AI and machine learning workloads.

Technical Details

Traditional data center networking relies heavily on vendor-specific CLIs, complex BGP/EVPN configurations, and manual provisioning. Netris replaces this legacy approach with a controller-based architecture that abstracts underlying switch hardware (from vendors like NVIDIA/Mellanox, Edgecore, and Dell) into a VPC-like interface. By leveraging open networking standards and Linux-based network operating systems, Netris allows infrastructure engineers to define network topologies, load balancing, and routing via Kubernetes CRDs or Terraform. For AI clusters, this means automated, declarative provisioning of the high-bandwidth, low-latency RDMA over Converged Ethernet (RoCEv2) fabrics required by distributed training jobs.

Why It Matters

The explosion of generative AI has created a massive supply-demand imbalance for compute, leading to the rise of specialized AI neoclouds (such as CoreWeave or Lambda). However, standing up thousands of GPUs is heavily bottlenecked by the physical network. GPU clusters are highly sensitive to network latency and packet loss; a single misconfigured switch can degrade an entire distributed training run. Netris attacks this bottleneck by turning bare-metal network configuration into declarative software. For an engineer building an AI cloud, this reduces the reliance on highly specialized network architects, replacing brittle manual setups with standard, version-controlled DevOps workflows.

What To Watch Next

Monitor Netris's hardware compatibility matrix with next-generation AI networking gear, particularly 800G Ethernet switches and NVIDIA's Spectrum-X platform. Additionally, it will be critical to see if top-tier AI neoclouds publicly adopt Netris as their default network control plane, or if they ultimately opt to build proprietary controllers in-house to squeeze out marginal performance gains at scale.